Uploaded image for project: 'OpenShift Autoscaling'
  1. OpenShift Autoscaling
  2. AUTOSCALE-491

Introduce OpenshiftNodePool to interface Karpenter NodePool

XMLWordPrintable

    • Icon: Story Story
    • Resolution: Unresolved
    • Icon: Undefined Undefined
    • None
    • None
    • None
    • None

      Similar to https://issues.redhat.com//browse/HOSTEDCP-2262 we want to introduce a new CRD called OpenshiftNodePool, which will render the regular upstream karpenter NodePool in the guest cluster, but with the ability to allow HCP to set default immutable values, hide fields that customers should not be able to touch, or even add our own openshift specific fields.

      Story 1: we would be able to set nodepool launch template requirements that prevent the launch of instancetypes with 2 CPUs, which is too little for an openshift node.

      Story 2: we may want to add our own openshift specific fields to the nodepool.

      Story 3: we may want to enforce that nodes have a label when they are created. It seems we actually need this for the `hypershift.openshift.io/nodepool-globalps-enabled: "true"` label.

      This story allows HCP to be more flexible with controlling the user's experience on a managed service with Karpenter.

      Acceptance criteria:

      1. OpenshiftNodePool CRD exists in the guest cluster
      2. VAP exists and is constantly reconciled to prevent modification or deletion of NodePool
      3. Changes from OpenshiftNodePool are reflected to NodePool

      I don't think there is a need right now to hide any fields, so reflecting the entire spec and status should be fine. Subject to change as we get more feedback.

              rh-ee-macao Max Cao
              rh-ee-macao Max Cao
              Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

                Created:
                Updated: